This application requests support for a two-year pilot study of memory in children with autism. Microscopic abnormalities have been found in the hippocampal complex, amygdala, and adjacent cortex within the limbic system of individuals with autism. Certain nonverbal learning and memory tasks have been shown to be sensitive to damage to those structures in studies with humans and nonhumans. We propose to use those tasks to evaluate basic learning and memory processes in children with autism whose severe language impairments preclude valid assessment by most traditional methods. The Specific Aims are to compare the performances of a sample of children with autism and those of a contrast group of nonautistic children with mental retardation on the following tasks: (1) Delayed nonmatching-to-sample (DNMS) with trial-unique stimuli. A novel stimulus (sample) is presented first on each trial, followed after a brief delay by presentation of that same stimulus with another novel stimulus (comparisons); a response to the comparison that does not match the sample is reinforced. Recognition memory is tested by manipulating the delay between sample removal and comparison presentation. This task is used widely to evaluate memory impairments in nonhumans and humans, and accurate performance at delays of more than 10 seconds appears to depend on an intact cortico-limbic system. (2) Delayed spatial alternation (DSA). This task represents another nonverbal test of memory that is sensitive to hippocampal damage. Subjects learn to alternate responding between two identical stimuli in different left-right positions with a brief delay between trials. Memory is challenged by increasing the delay. (3) Two-choice simple visual discrimination. For this task, two nonidentical stimuli are presented simultaneously on repeated trials. Responses to one stimulus (5+) are always reinforced, while responses to the other stimulus are not reinforced, until the subject responds exclusively to the 5+, Unlike DNMS, acquisition and retention of simple discriminations does not appear to be affected significantly by cortico- limbic damage.